I design software for people at work — the tools they open every day to get their jobs done. Not how do I get someone to book a ride, but how do I help someone do their job better. Lately, I’ve been most interested in how AI changes my work and those tools.
I like this work because it’s hard. The problems are big and dense. The users know more about their work than I do, and I have to catch up before I can design anything useful — the fifteen advocates I work with at Lyft will tell me, very quickly, when something on the screen is wrong. I learn something new on every project. The gap between a tool that helps and a tool that gets in the way is bigger than people think, and design is most of that gap.
The way we do that work is about to change. AI isn’t going to replace designers, but it is going to stop us from drawing every screen with a mouse. We’re moving toward writing the prompts and the setup around them, and reviewing what comes back.
I’ve been working this way in Claude Code at Lyft and with two startups. I’m writing prototypes engineers can actually run. The case studies go into how. This is the most consequential shift my job has had in a decade.